lecture 3 data mining primitives, languages, and system...
TRANSCRIPT
2006-3-12 Data Mining:Tech. and Appl. 1
Lecture 4aData Mining Primitives, Languages, and System Architectures
Zhou Shuigeng
March 12, 2006
2006-3-12 Data Mining:Tech. and Appl. 2
OutlineData mining primitives(原语): What defines a data mining task?
A data mining query language
Design graphical user interfaces based on a data mining query language
Architecture of data mining systems
Summary
2006-3-12 Data Mining:Tech. and Appl. 3
OutlineData mining primitives: What defines a data mining task?
A data mining query language
Design graphical user interfaces based on a data mining query language
Architecture of data mining systems
Summary
2006-3-12 Data Mining:Tech. and Appl. 4
Why Data Mining Primitives and Languages?
Finding all the patterns autonomously in a database? —unrealistic because the patterns could be too many but uninterestingData mining should be an interactive process
User directs what to be minedUsers must be provided with a set of primitives to be used to communicate with the data mining systemIncorporating these primitives in a data mining query language
More flexible user interaction Foundation for design of graphical user interfaceStandardization of data mining industry and practice
2006-3-12 Data Mining:Tech. and Appl. 5
What Defines a Data Mining Task ?
Task-relevant data
Type of knowledge to be mined
Background knowledge
Pattern interestingness measurements
Visualization of discovered patterns
2006-3-12 Data Mining:Tech. and Appl. 6
Task-Relevant Data
Database or data warehouse name
Database tables or data warehouse cubes
Condition for data selection
Relevant attributes or dimensions
Data grouping criteria
2006-3-12 Data Mining:Tech. and Appl. 7
Types of knowledge to be mined
Characterization
Discrimination
Association
Classification/prediction
Clustering
Outlier analysis
Other data mining tasks
2006-3-12 Data Mining:Tech. and Appl. 8
Background Knowledge: Concept Hierarchies
Schema hierarchyE.g., street < city < province_or_state < country
Set-grouping hierarchyE.g., {20-39} = young, {40-59} = middle_aged
Operation-derived hierarchyemail address: [email protected] < department < university < country
Rule-based hierarchylow_profit_margin (X) <= price(X, P1) and cost (X, P2) and (P1 - P2) < $50
2006-3-12 Data Mining:Tech. and Appl. 9
Measurements of Pattern Interestingness
Simplicitye.g., (association) rule length, (decision) tree size
Certaintye.g., confidence, P(A|B) = #(A and B)/ #(B), classification reliability or accuracy, certainty factor, rule strength, rule quality, discriminating weight, etc.
Utilitypotential usefulness, e.g., support (association), noise threshold (description)
Noveltynot previously known, surprising (used to remove redundant rules, e.g., Canada vs. Vancouver rule implication support ratio)
2006-3-12 Data Mining:Tech. and Appl. 10
Visualization of Discovered Patterns
Different backgrounds/usages may require different forms of representation
E.g., rules, tables, crosstabs, pie/bar chart etc.Concept hierarchy is also important
Discovered knowledge might be more understandable when represented at high level of abstractionInteractive drill up/down, pivoting, slicing and dicingprovide different perspectives to data
Different kinds of knowledge require different representation: association, classification, clustering, etc.
2006-3-12 Data Mining:Tech. and Appl. 11
Data mining primitives: What defines a data mining task?
A data mining query language
Design graphical user interfaces based on a data mining query language
Architecture of data mining systems
Summary
Outline
2006-3-12 Data Mining:Tech. and Appl. 12
A Data Mining Query Language (DMQL)Motivation
A DMQL can provide the ability to support ad-hoc and interactive data miningBy providing a standardized language like SQL
Hope to achieve a similar effect like that SQL has on relational databaseFoundation for system development and evolutionFacilitate information exchange, technology transfer, commercialization and wide acceptance
DesignDMQL is designed with the primitives described earlier
2006-3-12 Data Mining:Tech. and Appl. 13
Syntax(句法) for DMQL
Syntax for specification oftask-relevant data the kind of knowledge to be mined concept hierarchy specification interestingness measure pattern presentation and visualization
Putting it all together—a DMQL query
2006-3-12 Data Mining:Tech. and Appl. 14
Syntax: Specification of Task-Relevant Data
use database database_name, or use data warehouse data_warehouse_name
from relation(s)/cube(s) [where condition]
in relevance to att_or_dim_list
order by order_list
group by grouping_list
having condition
2006-3-12 Data Mining:Tech. and Appl. 15
Specification of task-relevant data
2006-3-12 Data Mining:Tech. and Appl. 16
Syntax: Kind of knowledge to Be Mined
CharacterizationMine_Knowledge_Specification ::=
mine characteristics [as pattern_name] analyze measure(s) (e.g. sum, count etc.)
DiscriminationMine_Knowledge_Specification ::=
mine comparison [as pattern_name] for target_class where target_condition{versus contrast_class_i where contrast_condition_i}analyze measure(s)
E.g. mine comparison as purchaseGroupsfor bigSpenders where avg(I.price) >= $100versus budgetSpenders where avg(I.price) < $100analyze count
2006-3-12 Data Mining:Tech. and Appl. 17
Syntax: Kind of Knowledge to Be Mined (cont.)
AssociationMine_Knowledge_Specification ::= mine associations [as pattern_name]
[matching <metapattern>]E.g. mine associations as buyingHabits
matching P(X:custom, W) ^ Q(X, Y)=>buys(X, Z)ClassificationMine_Knowledge_Specification ::=
mine classification [as pattern_name] analyze classifying_attribute_or_dimension
Other Patternsclustering, outlier analysis, prediction …
2006-3-12 Data Mining:Tech. and Appl. 18
Syntax: Concept Hierarchy Specification
To specify what concept hierarchies to useuse hierarchy <hierarchy> for <attribute_or_dimension>
We use different syntax to define different type of hierarchiesschema hierarchies
define hierarchy time_hierarchy on date as [date,month quarter,year]
set-grouping hierarchiesdefine hierarchy age_hierarchy for age on customer as
level1: {young, middle_aged, senior} < level0: alllevel2: {20, ..., 39} < level1: younglevel2: {40, ..., 59} < level1: middle_agedlevel2: {60, ..., 89} < level1: senior
2006-3-12 Data Mining:Tech. and Appl. 19
Concept Hierarchy Specification (Cont.)
operation-derived hierarchiesdefine hierarchy age_hierarchy for age on customer as {age_category(1), ..., age_category(5)} := cluster(default, age, 5) < all(age)
rule-based hierarchiesdefine hierarchy profit_margin_hierarchy on item as level_1: low_profit_margin < level_0: all
if (price - cost)< $50level_1: medium-profit_margin < level_0: all
if ((price - cost) > $50) and ((price - cost) <= $250)) level_1: high_profit_margin < level_0: all
if (price - cost) > $250
2006-3-12 Data Mining:Tech. and Appl. 20
Specification of Interestingness Measures
Interestingness measures and thresholds can be specified by a user with the statement: with <interest_measure_name> threshold = threshold_value
Example:with support threshold = 0.05with confidence threshold = 0.7
2006-3-12 Data Mining:Tech. and Appl. 21
Specification of Pattern Presentation
Specify the display of discovered patterns
display as <result_form>
To facilitate interactive viewing at different concept
level, the following syntax is defined:Multilevel_Manipulation ::= roll up on attribute_or_dimension
| drill down onattribute_or_dimension
| add attribute_or_dimension
| drop attribute_or_dimension
2006-3-12 Data Mining:Tech. and Appl. 22
Putting it all together: A DMQL query
use database AllElectronics_db use hierarchy location_hierarchy for B.addressmine characteristics as customerPurchasing analyze count% in relevance to C.age, I.type, I.place_made from customer C, item I, purchases P, items_sold S,
works_at W, branchwhere I.item_ID = S.item_ID and S.trans_ID =
P.trans_ID and P.cust_ID = C.cust_ID and P.method_paid = ``AmEx'' and P.empl_ID = W.empl_ID and W.branch_ID = B.branch_ID and B.address = ``Canada" and I.price >= 100
with noise threshold = 0.05 display as table
2006-3-12 Data Mining:Tech. and Appl. 23
Other Data Mining Languages & Standardization Efforts
Association rule language specifications
MSQL (Imielinski & Virmani’99)
MineRule (Meo Psaila and Ceri’96)
Query flocks based on Datalog syntax (Tsur et al’98)
OLEDB for DM (Microsoft’2000)
Based on OLE, OLE DB, OLE DB for OLAP
Integrating DBMS, data warehouse and data mining
CRISP-DM (CRoss-Industry Standard Process for Data Mining)
Providing a platform and process structure for effective data mining
Emphasizing on deploying data mining technology to solve business problems
2006-3-12 Data Mining:Tech. and Appl. 24
Data mining primitives: What defines a data mining task?
A data mining query language
Design graphical user interfaces based on a data mining query language
Architecture of data mining systems
Summary
Outline
2006-3-12 Data Mining:Tech. and Appl. 25
Designing Graphical User Interfaces Based on a Data Mining Query Language
What tasks should be considered in the design GUIs based on a data mining query language?
Data collection and data mining query composition
Presentation of discovered patterns
Hierarchy specification and manipulation
Manipulation of data mining primitives
Interactive multilevel mining
Other miscellaneous information
2006-3-12 Data Mining:Tech. and Appl. 26
Data mining primitives: What defines a data mining task?
A data mining query language
Design graphical user interfaces based on a data mining query language
Architecture of data mining systems
Summary
Outline
2006-3-12 Data Mining:Tech. and Appl. 27
Data Mining System Architectures
Coupling data mining system with DB/DW systemNo coupling—flat file processing, not recommendedLoose coupling
Fetching data from DB/DWSemi-tight coupling—enhanced DM performance
Provide efficient implement a few data mining primitives in a DB/DW system, e.g., sorting, indexing, aggregation, histogram analysis, multiway join, precomputation of some stat functions
Tight coupling—A uniform information processing environment
DM is smoothly integrated into a DB/DW system, mining query is optimized based on mining query, indexing, query processing methods, etc.
2006-3-12 Data Mining:Tech. and Appl. 28
Data mining primitives: What defines a data mining task?
A data mining query language
Design graphical user interfaces based on a data mining query language
Architecture of data mining systems
Summary
Outline
2006-3-12 Data Mining:Tech. and Appl. 29
SummaryFive primitives for specification of a data mining task
task-relevant datakind of knowledge to be minedbackground knowledgeinterestingness measuresknowledge presentation and visualization techniques to be used for displaying the discovered patterns
Data mining query languagesDMQL, MS/OLEDB for DM, etc.
Data mining system architectureNo coupling, loose coupling, semi-tight coupling, tight coupling
2006-3-12 Data Mining:Tech. and Appl. 30
ReferencesE. Baralis and G. Psaila. Designing templates for mining association rules. Journal of Intelligent Information Systems, 9:7-32, 1997.Microsoft Corp., OLEDB for Data Mining, version 1.0, http://www.microsoft.com/data/oledb/dm, Aug. 2000.J. Han, Y. Fu, W. Wang, K. Koperski, and O. R. Zaiane, “DMQL: A Data Mining Query Language for Relational Databases”, DMKD'96, Montreal, Canada, June 1996.T. Imielinski and A. Virmani. MSQL: A query language for database mining. Data Mining and Knowledge Discovery, 3:373-408, 1999.M. Klemettinen, H. Mannila, P. Ronkainen, H. Toivonen, and A.I. Verkamo. Finding interesting rules from large sets of discovered association rules. CIKM’94, Gaithersburg, Maryland, Nov. 1994.R. Meo, G. Psaila, and S. Ceri. A new SQL-like operator for mining association rules. VLDB'96, pages 122-133, Bombay, India, Sept. 1996.A. Silberschatz and A. Tuzhilin. What makes patterns interesting in knowledge discovery systems. IEEE Trans. on Knowledge and Data Engineering, 8:970-974, Dec. 1996.S. Sarawagi, S. Thomas, and R. Agrawal. Integrating association rule mining with relational database systems: Alternatives and implications. SIGMOD'98, Seattle, Washington, June 1998.D. Tsur, J. D. Ullman, S. Abitboul, C. Clifton, R. Motwani, and S. Nestorov. Query flocks: A generalization of association-rule mining. SIGMOD'98, Seattle, Washington, June 1998.